INDEX

Explanations

phrases related to the investigation or scrutiny of questionable or unethical behavior

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the significance or impact of events or situations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

lez

-0.72

oller

-0.72

upt

-0.69

ople

-0.68

erest

-0.68

ijah

-0.67

stall

-0.66

eeee

-0.66

audi

-0.66

rolet

-0.64

POSITIVE LOGITS

 evidenced

1.20

 opposed

1.13

piring

1.10

 well

1.06

pires

1.05

pired

1.04

 soon

1.02

pects

1.01

 regards

0.99

phy

0.97

Activations Density 0.194%

phrases related to the investigation or scrutiny of questionable or unethical behavior

phrases indicating the significance or impact of events or situations

No Comments

No Known Activations